Nishikawa plots

First, we plot the presence/absence data from the Nishikawa dataset for bluefin tuna. The positive sampling points are confined to April-June and July-September.

January-March

Note that for this season, there are no positive samples.

April-June

Most positive samples are south of Japan.

July-September

Not much positive sampling points. Those that are present are south of Japan.

October-December

No positive sampling points.

Base model (Model 1)

Then, we build the basic tuna model with all the predictors included. Note that the environmental predictors are mean values over 1956-1981.

  1. longitude
  2. latitude
  3. season
  4. temperature (\(^{\circ} C\))
  5. oxygen (\(mol \; m^{-3}\))
  6. pH
  7. chlorophyll (\(mol \; m^{-3}\))
  8. salinity (ppt)
  9. mixed layer thickness (m)
  10. nitrate (\(mol \; m^{-3}\))
  11. phosphate (\(mol \; m^{-3}\))
  12. ammonium (\(mol \; m^{-3}\))
  13. broad-scale thermal gradient (\(\Delta ^{\circ} C \; km^{-1}\))
  14. broad-scale salinity gradient (\(ppt \; km^{-1}\))
  15. eddy kinetic energy (\(m^{-2} \; s^{-2}\))
  16. depth (m)
  17. distance from nearest coast (km)
  18. probability of adult occurrence
## [1] "training AUC: 0.9995"
## [1] "testing AUC: 0.9977"

Then, we extrapolate for the rest of \(40^{\circ}N\)-\(40^{\circ}S\) and present seasonal distribution maps. The distribution maps are shown side-by-side with the Nishikawa maps.

January-March

I feel like this makes stuff up because there are no positive sampling points in this season.

April-June

July-September

October-December

I feel like this makes stuff up because there are no positive sampling points in this season.

Model without geographical location (Model 2)

## [1] "training AUC: 0.9979"
## [1] "testing AUC: 0.9932"

Again, each seasonal distribution map is shown side-by-side with its corresponding Nishikawa seasonal chart.

January-March

April-June

July-September

October-December

Increasing confidence of distribution maps.

For this section, we use Model 1 (full model). First, we build the \(10 \times 10\) grid.

January-March

For each season, we associate the \(10 \times 10\) grid with the \(1 \times 1\) grid cells. Then, we limit the area to \(10 \times 10\) grid cells with sampling points.

We only leave \(10 \times 10\) grid cells that have sampling points within a certain % area threshold. We first do this for a more conservative 25% threshold.

We also do it for a more liberal 10% threshold.

Then, we replicate this across the 3 other seasons…

April-June

July-September

October-December